同型机调度问题的混合ACO-SA优化算法研究

Research of Hybrid Ant Colony Optimization-Simulated Annealing Approach for Scheduling Identical Parallel Machines

针对同型机调度问题,提出一种蚁群-模拟退火两阶段优化算法.构造了问题域蚁群模型,运用蚁群算法展开全局搜索,通过自适应调整阈值改善空间探索与局部开采的平衡;为提高搜索精度,引入模拟退火算法,将蚁群算法的最好解作为其初始解,在邻域内进行精细搜索,利用其概率突跳特性有效避免算法陷入局部最优.实验结果表明混合算法具有稳定而优良的寻优能力.

This paper addresses a makespan minimization scheduling problem on identical parallel machines. An effective two-phase hybrid optimization strategy with ant colony optimization （ACO） and simulated armealing （SA） is proposed. Since ACO can produce better initial solutions, it is used to do global search in the problem space at first, by reconstructing the ant colony model and getting the balance of exploration and exploitation through adapting threshold. Then SA is introduced to improve the searching precision. The solutions of AC~） are used as good initial solutions in SA, which could do elaborate search in the neighbourhood and avoid trapping into kraal minima with its probability jump property. Computational results demonstrate that the hybrid one is very accurate and stable.